Using assumptions while curve fitting

Hello all,

I have a simple (I think!) question. I have an xy data set that I need to fit to a custom equation. x values range from 0 to 500 (they are concentrations, so negative values are unphysical/nonsensical), y values range from ~135 to 260.

The issue is this. The custom equation I need is a hyperbola. For x>0 the hyperbola the curve fitting tool solves for is excellent. However, I am left (in the curve fitting window) with a chunk of the hyperbola that is also for x<0. This piece of the function, which is nowhere near a data point, seems to be leading to an artificially low R^2 and other goodness-of-fit statistics.

Can I build an assumption into the curve fitting tool or command line fitting procedure where my custom equation assumes x>=0 and so yields reasonable goodness-of-fit values?

Thank you sincerely for any assistance you can provide!

Sam

[Walter Roberson]

Which routine are you using to do the fitting?

Answer 1

[Shashank Prasanna]

Curve fitting tools will fit the equation you are trying to model. This equation may very well only be called for values of x < 0.

Since R^2 is computed purely by using your data - which I assume does not have x < 0 values, then what you see there is a purely visual artifact. You can continue to use the fitted model for your x > 0 only.

If this does not help, could you provide details on what is the exact equation you are trying to fit as well as how the data looks.

[James]

Shashank, thank you so much, I did not realize that. This makes my job much easier. Thanks again!

Sam